Application of analytical redundancy management to shuttle crafts

This paper presents a description of the analytical redundancy management section of an experiment, under development for the space shuttle. The proposed experiment uses a modification of the bank of filters approach, described by Montgomery and Price earlier (AIAA J. of Aircraft, Vol. 13, No. 2, pp. 76-82, Feb. 1976). It is performed by injecting failures into simulated sensor data, generated in real-time or batch simulations of shuttle reentry and on-orbit flight phases. The previous analysis contained a bank of extended Kalman filters running in parallel. In the current study, these filters have been replaced by UD factorized filters. This was done to facilitate a microelectronic implementation. A comparative study of the previous, and current setup, through simulation of the microelectronics is being conducted. Data, obtained using the simulations, are presented. These data indicate that both the Kalman and UD factorized filters suffer from word bit size limitations to about the same extent principally because of modeling error. The principle advantage of the UD factorized filter therefore lies in the stability for serial measurement processing.